A note on abscissas of Dirichlet series (Q2317595)

This paper is concerned with the abscissae of convergence of Banach-valued Dirichlet series. Given a Banach space \(X\) the authors use \(\mathcal{D}(X)\) to denote the spaces of all Dirichlet series, \(\sum_na_nn^{-s}\), with values in \(X\). The abscissa of absolute (resp. uniform) convergence \(\sigma_a(D)\) (resp. \(\sigma_u(D)\)) of a Dirichlet series \(D\) is defined as the infimum of \(\sigma\) such that \(D\) converges absolutely (resp. uniformly) on \(\{s:\mathrm{Re}(s)> \sigma\}\). Given a Banach space of Dirichlet series \(\mathcal{X}(X)\) the abscissa \(\sigma_{\mathcal{X}(X)}\) is defined as \(\inf\{\sigma :\sum_n{\frac{a_n}{n^\sigma}} n^{-s} \text{ converges in } \mathcal{X}(X)\}\). A Banach space of Dirichlet series, \(\mathcal{X}(X)\), is said to be admissible if there are \(C_1,C_2>0\) so that \[ C_1\sup_n\|a_n\|\le \left\|\sum_n a_nn^{-s}\right\|_{\mathcal{X}(X)}\le C_2\sum_n\|a_n\| \] for all \(\sum_na_nn^{-s}\) in \(\mathcal{X}(X)\). For admissible Banach spaces, the authors obtain an upper bound for \(\sigma_{\mathcal{X}(X)}(D)\) in terms of the partial sums of \(D\). Given a Banach space \(X\) and \(1\le p\le \infty\) the authors use \(\mathcal{H}_p(X)\) to denote the image of the Hardy space \(H_p(X)\) under the Bohr transform. They prove that \(\sigma_{\mathcal{H}_p(X)}(D)=\sigma_{\mathcal{H}_q(X)}(D)\) for every \(D\) in \(\mathcal{D}(X)\) and \(1\le p \le q<\infty\). For an admissible space of scalar valued Dirichlet series \(\mathcal{X}\) and a Banach space \(X\) the space \(\mathcal{X}^{\mathrm{weak}}(X)\) is defined as all \(D=\sum_na_nn^{-s}\) in \(\mathcal{X}(X)\) such that \(\sup_{\phi\in X^*}\|\sum_n\phi(a_n)n^{-s}\|_{\mathcal{X}(X)}<\infty\). It is shown that \(\mathcal{X}^{\mathrm{weak}}(X)\) is admissible and that \(\sigma_{\mathcal{X}^{\mathrm{weak}}(X)}(D)=\sup_{\|\phi\|\le 1}\sigma_{\mathcal{X}(X)} \phi (D)\). For a Dirichlet series \(D=\sum_na_nn^{-s}\) the abscissa of unconditional convergence, \(\sigma_{\mathrm{unc}}(D)\), is defined as the infimum of \(\sigma\) so that \(\sum_na_nn^{-s}\) converges unconditionally on \(\{s:\mathrm{Re}(s)> \sigma\}\). The authors show that, for a Banach space \(X\): \begin{itemize}\item[(i)] \(\sup\{\sigma_a(D)-\sigma_{\mathrm{unc}} (D): D\in \mathcal{D}(X)\}=1-\frac{1}{\mathrm{cot}(X)}\), where \(\mathrm{cot}(X)\) is the cotype of \(X\), \item[(ii)] \(\sup\{\sigma_{\mathrm{unc}}(D)-\sigma_{u}(D): D\in \mathcal{D}(X)\} =\frac{1}{2}\); \item[(iii)] For any infinite-dimensional Banach space \(X\) and \(0\le \alpha\le \frac12\), there is an \(X\)-valued Dirichlet series \(D\) such that \(\sigma_u(D)=0\), \(\sigma_{\mathrm{unc}}(D)=\alpha\) and \(\sigma_{a}(D)=1-\frac{1}{\mathrm{cot} (X)}\). \end{itemize}